新型混沌粒子群混合优化算法

摘要

为避免粒子群算法陷入局部最优、早熟收敛,提出了一种新型的混沌粒子群混合优化算法.利用混沌映射初值敏感性、遍历性特点,随机初始化一个粒子,并通过混沌映射得到多个粒子的初始值,改变初始粒子群的提取过程.利用混沌映射扩大初始粒子群,得到寻优粒子群,使得粒子群在搜索的过程中,种群数量变大,有利于全局寻优,而种群粒子多样化,有利于跳出局部极值.经典的测试函数仿真表明,改进的粒子群算法极大提高了粒子群的寻优精度和寻优效率,增加了粒子的全局寻优能力,具有更为广泛的应用场景.%To avoid falling into premature and local convergence,a new chaotic particle swarm hybrid optimization algorithm is put forwards.Firstly,the algorithm using the characteristic of ergodicity of chaos movement and the character of being highly sensitivity to the original value,randomly initializing a particle,obtaining many particles' initial value through chaotic map,change the original particle swarm extracting process;Then,expand the amount of the original particle swarm to get optimizing particle swarm,therefore,in the searching process,the particle swarm get bigger and be helpful for global optimization;Finally,it can make the optimizing particle swarm diversity,making local optimization easier.The simulations of four classic test function show that the improved particle swarm algorithm behaves better than the basic particle swarm optimization algorithm in convergent speed and accuracy.It greatly improves the ability of global optimization,having wider application scope.